In his first book, Wittgenstein presented a unified ontology for "everything that is the case." The totality of facts, he claimed, can be stated clearly in language or logic, and "Whereof one cannot speak, thereof one must be silent." That book was the foundation for formal semantics in the 20th century. Yet those precise formalisms were also brittle and inflexible. In his later philosophy, Wittgenstein replaced the monolithic logic and ontology of his first book with an open-ended family of language games. In principle, that framework could accommodate any kind of language, but programmers despaired of implementing such a disorganized collection. This talk compares Wittgenstein's ideas to trends in artificial intelligence. Monolithic frameworks similar to his early approach have had limited success, but his later ideas suggest promising ways of developing, relating, and using dynamically evolving ontologies. These methods have important implications for AI, computational linguistics, and the Semantic Web.
John F. Sowa spent thirty years working on research and development projects at IBM and is a co-founder of VivoMind Intelligence, Inc. He has been working on novel methods for combining logic-based symbolic processing of conceptual graphs with continuous numeric computations of knowledge signatures. This combination has made major improvements in the efficiency of algorithms for communication, reasoning, and natural language processing.
Dynamic Ontology, A Wittgensteinian Method of Relating Language to the World Wednesday 8 October 2008, 4.00pm